CA2709611C - Non-destructive testing, in particular for tubes during manufacture or in the finished state - Google Patents

Non-destructive testing, in particular for tubes during manufacture or in the finished state Download PDF

Info

Publication number
CA2709611C
CA2709611C CA2709611A CA2709611A CA2709611C CA 2709611 C CA2709611 C CA 2709611C CA 2709611 A CA2709611 A CA 2709611A CA 2709611 A CA2709611 A CA 2709611A CA 2709611 C CA2709611 C CA 2709611C
Authority
CA
Canada
Prior art keywords
tube
imperfection
ultrasonic
echoes
arrangement
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CA2709611A
Other languages
French (fr)
Other versions
CA2709611A1 (en
Inventor
Frederic Lesage
Nidia Alejandra Segura Rodriguez
Bernard Bisiaux
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
V&M France
Original Assignee
V&M France
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Priority to FR07/09045 priority Critical
Priority to FR0709045A priority patent/FR2925690B1/en
Application filed by V&M France filed Critical V&M France
Priority to PCT/FR2008/001751 priority patent/WO2009106711A2/en
Publication of CA2709611A1 publication Critical patent/CA2709611A1/en
Application granted granted Critical
Publication of CA2709611C publication Critical patent/CA2709611C/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • GPHYSICS
    • G01MEASURING; TESTING
    • G01NINVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N29/00Investigating or analysing materials by the use of ultrasonic, sonic or infrasonic waves; Visualisation of the interior of objects by transmitting ultrasonic or sonic waves through the object
    • G01N29/04Analysing solids
    • G01N29/11Analysing solids by measuring attenuation of acoustic waves
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01NINVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N29/00Investigating or analysing materials by the use of ultrasonic, sonic or infrasonic waves; Visualisation of the interior of objects by transmitting ultrasonic or sonic waves through the object
    • G01N29/04Analysing solids
    • G01N29/06Visualisation of the interior, e.g. acoustic microscopy
    • G01N29/0609Display arrangements, e.g. colour displays
    • G01N29/0645Display representation or displayed parameters, e.g. A-, B- or C-Scan
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01NINVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N29/00Investigating or analysing materials by the use of ultrasonic, sonic or infrasonic waves; Visualisation of the interior of objects by transmitting ultrasonic or sonic waves through the object
    • G01N29/22Details, e.g. general constructional or apparatus details
    • G01N29/26Arrangements for orientation or scanning by relative movement of the head and the sensor
    • G01N29/27Arrangements for orientation or scanning by relative movement of the head and the sensor by moving the material relative to a stationary sensor
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01NINVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N29/00Investigating or analysing materials by the use of ultrasonic, sonic or infrasonic waves; Visualisation of the interior of objects by transmitting ultrasonic or sonic waves through the object
    • G01N29/44Processing the detected response signal, e.g. electronic circuits specially adapted therefor
    • G01N29/4445Classification of defects
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01NINVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N29/00Investigating or analysing materials by the use of ultrasonic, sonic or infrasonic waves; Visualisation of the interior of objects by transmitting ultrasonic or sonic waves through the object
    • G01N29/44Processing the detected response signal, e.g. electronic circuits specially adapted therefor
    • G01N29/4481Neural networks
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01NINVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N2291/00Indexing codes associated with group G01N29/00
    • G01N2291/01Indexing codes associated with the measuring variable
    • G01N2291/011Velocity or travel time
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01NINVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N2291/00Indexing codes associated with group G01N29/00
    • G01N2291/02Indexing codes associated with the analysed material
    • G01N2291/023Solids
    • G01N2291/0234Metals, e.g. steel
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01NINVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N2291/00Indexing codes associated with group G01N29/00
    • G01N2291/02Indexing codes associated with the analysed material
    • G01N2291/028Material parameters
    • G01N2291/02854Length, thickness
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01NINVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N2291/00Indexing codes associated with group G01N29/00
    • G01N2291/04Wave modes and trajectories
    • G01N2291/044Internal reflections (echoes), e.g. on walls or defects
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01NINVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N2291/00Indexing codes associated with group G01N29/00
    • G01N2291/10Number of transducers
    • G01N2291/105Number of transducers two or more emitters, two or more receivers
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01NINVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N2291/00Indexing codes associated with group G01N29/00
    • G01N2291/26Scanned objects
    • G01N2291/263Surfaces
    • G01N2291/2634Surfaces cylindrical from outside

Abstract

Device forming a tool for assisting the exploitation, for the control destructive of iron and steel products, intended to information about possible imperfections of the product, from signals back by ultrasonic sensors transmitters, receiving ultrasonic sensors forming an arrangement of selected geometry, mounted in ultrasonic coupling with the product via a liquid medium, with relative movement of rotation / translation between the tube and the arrangement of transducers, said operating tool being characterized in that includes: a converter (891; 892) capable of isolating selectively a numerical representation of possible echoes in designated time windows, depending on the relative rotation / translation movement, said representation comprising the amplitude and flight time of at least one echo, and to generate a parallelepiped 3D graph, a transformation block (930) capable of generating a 3D image (901; 902) possible imperfections in the tube from the 3D graph and a base of data, a filter (921; 922), able to determine, in the images (901; 902), areas of presumed imperfection (Zcur), as well as that properties of every presumed imperfection, and an output stage configured to generate a compliance signal or not conformity of a product.

Description

Non-destructive testing, in particular for tubes being manufactured or in the finished state.
The invention relates to the non-destructive testing of materials, particularly for tubes Manufacturing.
Various proposals are already known, to which we will return, tending use neural networks as part of non-destructive testing of materials.
But this existing is not likely to operate in an industrial environment, on equipment already in service, in real time, and while allowing a classification in flight of imperfections according to their nature, so that we can remedy quickly has a problem occurred in the production phase.
An object of the invention is to improve the situation towards a system which can:
- be used in an industrial environment and easily installed on equipment already existing in this environment, - be used in real time, that is to say give a quick diagnosis, particular to sufficient speed not to slow down the overall production speed, and - allow a classification of imperfections according to their nature, to from a weak amount of information, so as to know their seriousness and to allow the determination of the technical cause causing the imperfection and thus remedy quickly to the problem in the production phase.
According to a first aspect, it is proposed a device that forms a tool to help operation, for the non-destructive testing of tubes (or other products steel) in progress and out of production. Such a tool is intended to draw information on possible imperfections of the product. Ultrasonic emitter sensors are excited

2 selectively according to a chosen law of time. Feedback signals are captured by receiving ultrasonic sensors forming a chosen geometry arrangement, mounted in ultrasonic coupling with the tube via a liquid medium.
Finally, there is generally a relative movement of rotation / translation between the product and the arrangement of transducers.
The proposed farm support tool includes:
a converter capable of selectively isolating a representation digital possible echoes in designated time windows, depending on the movement relative rotation / translation, and to draw an image of imperfections possible in the product, said representation comprising the amplitude and the flight time from to minus one echo, and generate a parallelepiped 3D graph, - a transformation block capable of generating a 3D image of imperfections possible in the tube from the 3D graph and a database, - a filter, able to determine, in the images, areas of imperfection alleged, as well as properties of every presumed imperfection, an output stage configured to generate a signal of conformity or not conformity of a product.
The invention can also be placed at the level of a control device destructive of tubes (or other iron and steel products) in process or at the time of manufacture, who includes:
an arrangement of ultrasonic transducers of selected geometry mounted in coupling ultrasound with the tube via a coupling medium, with movement relative rotation / translation between the tube and the arrangement of transducers circuits for selectively exciting these transducer elements according to a law of chosen time, and to collect the return signals they are capturing, and - an operating aid tool as above.
Another aspect of the invention is expressed in the form of a non-controlled control method.

destruction of tubes (or other iron and steel products) on or off of manufacture, comprising the following steps:

3 at. provide an arrangement of ultrasonic transducers of geometry selected, mounted in ultrasonic coupling with the tube via a medium coupling, with relative movement of rotation / translation between the tube and the arrangement of transducers, b. selectively excite these transducer elements according to a time law selected, vs. collect the feedback signals they capture, in order to analyze selectively these return signals, to draw information about any imperfections of the tube, said information including the amplitude and the flight time from to less an echo and generate a parallelepipedic 3D graph.
d. selectively isolate a digital representation of possible echoes in designated time windows, depending on the relative movement of rotation / translation, and draw a 3D image of possible imperfections in the tube from the parallelepiped 3D graph and a database, e. generate a signal of conformity or nonconformity of a product.
Step e can include:
el. filter the images according to selected filtering criteria, in order to determine areas of presumed Zcur imperfection, as well as properties of each presumed imperfection, e2. form digital work inputs, from an excerpt from imagery corresponding to an area of presumed imperfection Zcur, of properties of the presumed imperfection in the same zone, resulting from the filter; and data from context, e3. apply the entries thus formed to at least one arrangement of the kind circuit neural, e4. digitally process the output of the layout of the circuit genre neuronal selected decision criteria, to make a decision and / or an alarm, and e5. discard and mark tubes decided non-compliant by step e4.
Other aspects, characteristics and advantages of the invention will appear in the examination of the following detailed description of some embodiments not limiting, as well as annexed drawings in which:

4 FIG. 1 is a schematic perspective view of a tube, having imperfections or so-called standard defects;
FIG. 2 is a schematic side view illustrating an example installation type "rotating head control" on a tube at the outlet of manufacture;
FIGS. 3A to 3C are details of different types of measurement thick and control of longitudinal and transverse imperfections;
FIG. 4 is a schematic diagram of the electronics associated with a ultrasonic sensor in non-destructive testing in a conventional installation;
FIGS. 5A and 5B are an end view and a side view of a type particular of non-destructive control cell, commonly referred to as "rotating head" and represent schematically;
FIG. 6 shows the complexity of the ultrasonic paths encountered in a tube on a simple example;
FIGS. 6A and 6B are schematic time diagrams of signals ultrasound, for a sensor under oblique incidence, and for a sensor under impact normal (perpendicular), respectively;
FIG. 7 is a graph showing a conventional representation of the selectivity of a control installation;
FIG. 8 is a schematic diagram of the electronics associated with a sensor ultrasound in non-destructive testing in an installation example susceptible to implement the invention;
FIG. 8A is a more detailed block diagram of part of the Figure 8;
FIG. 8B is another more detailed block diagram of a portion of Figure 8;
FIG. 9 is a schematic screen copy illustrating images ultrasonic digitized potential imperfections in a tube;
FIG. 9A is a screen shot in another orientation;
FIGS. 10A to 10D are schematic representations of different Types imperfections according to the API classification (American Petroleum Institute) and who constitute the output data of the neural network tending to determine the type of imperfection;
- Figure 11 is a more detailed block diagram of another part of Figure 8;
FIG. 11A is a detailed view of the transformation block of FIG.
11;

FIG. 12 is a sequential diagram illustrating the treatment imperfections successive potentials in an image;
FIG. 13 is a block diagram of a filter system;
FIG. 14 is the block diagram of a neural network assembly tending to determine the type of imperfection in a tube;
FIG. 15 is the block diagram of a neural network assembly tending to determine the degree of severity of an imperfection in a tube;
FIG. 16 is the block diagram of the neuron model;
FIG. 17 is an example of a transfer function of a neuron elementary; and FIG. 18 is a general diagram of an installation for the detection of defects on different types of sensors.
The drawings contain elements of a certain character. They will be able no only serve to better understand the present invention, but also contribute to its definition, if any.
In the rest of this text, an ultrasonic sensor may be designated indifferently by the terms sensor, or probe or transducer, well known of the skilled person.
Neural circuits The use of neural networks as part of the nondestructive testing of materials has been the subject of many publications, most of the time quite theoretical, which one will now consider.
The article "Localization and Shape Classification of Defects using the Finite Element Method and the Neural Networks "by ZAOUI, MARCHAND and RAZEK (NDT.NET-AUGUST 1999, vol. IV n abbreviated 8) formulates proposals in this area.
However, these proposals are made in the context of manipulations in laboratory, and the application described does not allow an implementation online, in the middle industrial. In in addition, only eddy current detection is processed, which is often insufficient.

The article "Automatic Detection of Defect in Industrial ultrasound images using a neural Network of Lawson and Parker (Proc, of Symposium on Lasers, Optics, and Vision for Productivity in Manufacturing I (Vision Systems: Applications), June 1996, Proc. of SPIE vol. 2786, pp. 37-47, 1996), describes the application of image processing and neural networks to the interpretation of what is called "TOFD scan". The method TOFD (Time of Flight Diffraction) consists in locating the positions of the sensor ultrasound where we can observe a diffraction of the beam on the edges of imperfection, which subsequently allows to size the imperfection.
This method is difficult to adapt to non-destructive testing equipment already existing, especially in industrial settings.
The article "Shape Classification of Flaw Indications in 3-Dimensional ultrasonic Images "
Dunlop and McNab (IEE Proceedings - Science, Measurement and Technology -july 1995 - Volume 142, Issue 4, p. 307-312) concerns the diagnosis in terms of corrosion pipeline. The system allows deep non-destructive testing and allows a study in three dimensions and in real time. However, the system is very slow.
This makes its use in an industrial environment relatively difficult.
The article "Application of neuro-fiizzy techniques in ultrasonic oil pipelines nondestructive testing "by Ravanbod (NDT & E International 38 (2005) p 643-653) suggests that imperfection detection algorithms can be improved by the use of fuzzy logic elements, mixed with the neural network.
However, The studied techniques also concern the inspection of imperfections of pipeline and a diagnosis on corrosion imperfections.
DE 42 01 502 C2 describes a method for creating a signal intended for a network of neurons, but provides little or no lessons on interpretation of the results, in terms of diagnosis. In addition, once again, only the detection by Eddy currents are treated.

Japanese Patent Publication No. 11-002626 relates to the detection imperfections longitudinal only, and only by eddy currents.
Japanese Patent Publication No. 08-110323 is content with a study in frequency signals obtained by ultrasound.
Japanese Patent Publication No. 2003-279550 describes a program for Do the difference between a zone qualified as healthy and a bad zone of a product using a network of neurons. This program does not go further, and does not allow the classification or the location of imperfections. Consequently, the application of this program can frequently lead to rejection of considered good if the results were interpreted by an operator human.
Non destructive testing of tubes The following detailed description is given primarily in the context of the control no destructive tubes at the output of manufacture, without limitation.
As shown in FIG. 1, the imperfections in a tube T can be distinguished according to their position. Thus, surface imperfections, internal or external, include longitudinal imperfections LD, and circumferential imperfections (or transverse or transverse or transverse) CD and oblique imperfections or inclined ID
; by different sensor arrangements, we try to detect them as soon as that they extend over a length and depth defined by the standards or specifications or customer specifications (for example a value of length imperfection quoted in the standards is 1/2 inch, or about 12.7 mm with a depth approximately 5% of the thickness of the controlled product). We are also interested in imperfections in the wall, that is to say in the MD mass (not visible in Figure 1), who often correspond to inclusions or dissensions, which one seeks to detect at the same time that a measurement of thickness is made. Ultrasonic beams are divergent representations in Figure 1 to understand the detection imperfections.
In practice, they will rather converge, as we will see.

Conventionally, in non-destructive ultrasonic testing, it is used one of the three types of installations: so-called "rotating head" installations, the installations so-called "rotating tube", and encircling sensor installations multielements, all well known to those skilled in the art. In the case of use of sensors operating in electronic scanning, the relative rotation -tube / sensors is virtual.
As used here, the expression "relative movement of rotation / translation between the tube and the arrangement of transducers "covers the case where the relative rotation is virtual.
In Figure 2, the non-destructive control machine with rotating head includes a ultrasonic device itself, mounted on a water chamber or "box with water "100, that crosses the tube T at the speed v = 0.5 meters per second, for example.
The sensors or ultrasonic probes emit longitudinal waves in the water. A
given sensor for example works at 1 or some MHz. He is excited repetitively by pulses, chosen waveform, at a rate (or frequency) of recurrence Fr which is of the order of a few kHz or tens of kHz, for example 10 kHz.
In addition, an ultrasound transducer has:
a radiation of near field, practically parallel, in a zone called of Fresnel, seat of many interferences, whose length in the axis of the beam is N = 0.25 D2 / 1 where D is the diameter of the active pellet of the transducer, and k its length Wave of work, and - far-field radiation, in an area called Fraunhofer, according to a beam diverging angle 2 a, with sin a = 1.22 1 / D
FIGS. 3A, 3B, 3C show converged sensors using means a concave (ultrasonic) lens, as commonly used in applications control of the tubes. The Fraunhofer area is preferably used less disrupted.

So, for sensors such as Pli and P12, the ultrasound beam, which is in general focused, extends in the vicinity of a plane perpendicular to the axis of the tube T.
detection is therefore substantially in cross section. Their roles are:
- or their beam is also perpendicular to the axis of the tube T in the section right, and they serve for the measurement of thickness (for example P1, FIG. 3A); we speak then "right feel";
- or their beam is incident on the axis of the tube T, in cross-section, and they serve to detect longitudinal imperfections (eg P11, Figure 3B). In that case, the angle of incidence in the cross-section is preferably selected for not cause in the tube as transverse or shear ultrasonic waves, considering characteristics of the water / metal interface of the tube (in principle water / steel).
We are planning generally two P1 and P12 sensors, of opposite incidences with respect to the axis of tube (Figure 2).
The machine also includes sensors such as P21 and P22 which, on the other hand, the ultrasound beam, which is also focused as a rule, extends to neighborhood of a plane passing through the axis of the tube, but incident with respect to the plane perpendicular to the axis of the tube T (see sensor P21, Figure 3C). In this case, the angle incidence by relative to the plane perpendicular to the axis of the tube is preferably chosen for generate in the tube only transverse ultrasonic waves or shear given the characteristics of the water / metal interface of the tube (in principle water / steel).
These sensors are used to detect transverse imperfections. We are planning usually two P21 and P22 sensors, of opposite incidences relative to the plane perpendicular to the axis of the tube (Figure 2).
The control of imperfections is usually done by focusing the beam.
Point of focus is measured against the "jump", which corresponds to the first path to go-ultrasound back in the thickness of the tube. So, the sensor of the figure 3A is focused at half-leap, while the sensors of Figures 3B and 3C are focused to three quarters of a leap. In addition, the control of external imperfections made generally leaping, and that of imperfections internal to the half-leap.

We note Ta the time of presence required for the probe to receive correctly the return of the ultrasound beam, representative of a possible imperfection. This time Ta depends on the sum of the two following times:
- on the one hand the round trip propagation delay of ultrasonic waves longitudinal, on the height of the "water column" present between the probe and the tube, on the journey of ultrasound, on the other hand, the propagation time of transverse ultrasonic waves, such as that it is required inside the tube to do the nondestructive testing itself.
This time depends mainly on a choice of the desired number of reflections from the waves transversely inside the wall of the tube.
Classically, the feelers are rotated about the axis of the tube, by means not shown, at a speed T of the order of several thousand tours by minute, (6000 rpm for example). In the case also known to the man of job where it is the tube which is rotated while the feelers are not not trained in rotation (so-called "rotating tube" installation) the speed of rotation of the tube is from the order of a few tens to a few thousand revolutions per minute.
One can call "cell" each set sensor - transmission medium (water) - tube.
For a cell, the beam opening Od of the feelers to ultrasound in detection. An opening can be defined with two components (Figure 1), one Odl in the right section of the tube, the other Od2 in the passing plane by the axis of tube and probe.
The setting of the installation (depending on the speed of rotation, the speed of scrolling, dimensions Odl and 0d2 and the number of probes) must guarantee a ultrasonic beam scanning of all surfaces and the volume of the tube to control.
It should be noted that certain standards or specifications or specifications client impose a recovery of swept areas.

The analysis time Ta is therefore defined by a compromise between:
the frequency (or frequency) of recurrence Fr, - in cross-section of the tube, the speed of rotation co, taking into account opening up Odl detection of ultrasonic probes (that is, taking into account the rotation of sensors, the Odl component of the beamwidth must allow a time of presence of the imperfection in front of the sensors which is at least equal to Ta), - along the tube, the speed of travel v of it, taking into account opening up 0d2 detection of an ultrasonic probe, and the number of NFi dedicated probes to the same function Fi (which therefore constitute a group of feelers), on the periphery of tube (in other words, given the advance of the tube, the 0d2 component of the opening beam should allow a time of presence of the imperfection in front of the sensor (or the group of sensors) which is at least equal to Ta).
- the number of probes dedicated to the same role (ie to the same function) , and the propagation times of the waves as defined above.
Typically, the machine typically comprises in all two sensors such as Fold, P12 for the control of imperfections of type LD, and possibly ID, two sensors such as P21, P22 for the control of CD-type imperfections, to which added in principle a sensor such as Pl, for the measurement of the thickness of the product and the control imperfections of MD type. Each sensor can be in fact a group of sensors working together, as we will see.
The machine has, in an integrated way or separately, an excitation electronics and of detection associated with each of the sensors. It includes (Figure 4) a transmitter pulses, for example 250 volts for the excitation of the probe PO mounted on the water box 100. As an integral part of the control system destructive, the ultrasonic probe PO, here transmitter / receiver, receives the echoes consecutive to this excitation. Lines 700 and 710 respectively transmit the pulse of excitement and the signal at the terminals of the probe to an amplifier 73.
The output of amplifier 73 is for viewing by the operator and / or driving a sorting machine, able to discard (downstream) non-compliant tubes.

The visualization is for example performed by an oscilloscope 750, which receives as signal the output of the amplifier 73, and as a time base 752 a signal one floor timing 753 from the transmitter 70. A threshold stage 754 avoid the blindness of the oscilloscope at the moment of the emission pulse.
Another output of amplifier 73 goes to a processing stage of signal 760. This Treatment typically includes straightening, smoothing, and filtering.
He is followed a detection stage or selector 762, capable of isolating echoes significant, known way. In detection of imperfection, it is the presence of an echo, with his amplitude or its duration (therefore its energy), which are significant, in some niches temporal, essentially the half-leap and the leap. For detection in thickness, we verifies that the equivalent-distance of the time difference between the bottom echoes respective corresponds to the desired thickness of the tube. The anomalies detected according to these criteria can be used to emit an alarm at 764, and / or to drive a controller 766 of sorting who evacuate the non-compliant tubes, marking them according to the anomaly or anomalies detected.
Materially in the case of a rotating head installation (Figures 5A and 5B), the cell further comprises, on a mechanical support 80, the water box 100, which lodge one PO sensor assembly, with a 701 connection, which connects the 700 and 710 lines of the FIG. 4 is for example three bearings 81 to 83 for centering the tube T.
According to the known technique (machine sold for example by the German company GE
NUTRONIK formerly NUKEM), the PO sensor assembly includes sensors which rotate at a few thousand revolutions / minute around the tube. We can also use a plurality of sensors distributed in a ring around the tube. The ring comprises for example 6 sectors of 128 ultrasonic sensors, distributed around the periphery.
The sectors of sensors are alternately slightly shifted in the direction of the axis of the tube. This makes it possible to have an overlap between two sectors of sensors consecutive, and also reduces the problems interference.
Interference occurs when a given sensor receives echoes due to shot made on another sensor.

To this is added a bench (not shown) for guiding the tube upstream and in downstream of non-destructive checkpoint, to properly position the tube that runs in continuous, by compared to ultrasonic sensors.
Non-destructive testing should be done on the entire periphery of the tube. But he is also essential that this control follows the linear velocity v of the tube in Release manufacturing. So we come to a compromise between the linear velocity v of tube, the rate (or frequency) of recurrence Fr, the analysis time Ta, the opening of job Od of the ultrasonic probe in detection, and the speed of rotation 6), the number of sensors ensuring the same function and the speed of propagation of the ultrasonic waves.
It is also desirable that the same installation can work on a whole range of tube diameters (and also tube thicknesses), covering the range production. It is then common to provide several values of speed rotation 6) and of the frequency of recurrence Fr, values which one selects according to diameter of the tube to be treated.
Finally, note that any manufacturing change involves a new setting of the ultrasonic angles of attack of each sensor on the periphery of the tube.
This delicate operation, performed manually, usually takes the order of a half-hour, time during which the production of tubes is stopped. Such are the conditions in which the non-destructive testing is currently carried out by ultrasonic tubes, or other profiled products and / or thin-walled, in Release manufacturing.
In the field of non-destructive ultrasonic testing, we often use the following terminology:
- scan (or scan) means a sequence of relative positions tube / sensor, - increment means the scan step (inversely proportional to the frequency of recurrence or frequency of ultrasound shots), - Ascan means the graph of the electrical voltage measured at the terminals of a sensor ultrasound, with the flight time on the abscissa and the ordinate one representation of the electrical voltage, also called ultrasonic amplitude, Bscan denotes an image relative to a given value of the increment, with, in abscissa, the scan corresponding to the ultrasonic firing, possibly expressed in degree angle of the sensor relative to the part to be inspected, ordinate the time of flight, and each point the ultrasonic amplitude converted to grayscale or color, - Echodynamic means a graph with the abscissa indicating the shot ultrasonic and in ordinate the maximum amplitude recorded in a temporal selector of the Ascan for the corresponding shot, - Cscan indicates an image with the abscissa and ordinate the position equivalent in a plane space of the point of firing of the ultrasonic wave and representing, converted to gradient of gray, the maximum ultrasonic amplitude for this shot recorded in the selector considered time of the Ascan ("amplitude of the image"). In the case of a tube, a point of the abscissa of the Cscan corresponds to a position on the length of the tube and a point of the ordinate corresponds to a position on the circumference of the tube. In the case a flat product, a point of the abscissa of the Cscan corresponds to a position on the length of the flat product and a point of the ordinate corresponds to a position on the width of the flat product.
Moreover, the plaintiff uses in the rest of the description the following terms:
- parallelepipedic 3D Bscan which designates a 3D representation comprising in in addition to the position of the sensor on the axis of the tube, the representation being considered as gross, the shape of the tube does not appear, - Reduced 3D Bscan which designates a parallelepiped 3D Bscan limited to a zoned with an ultrasonic indication of a probable fault after filtering, - Bscan 3D tube that has the same dimensions as the Bscan 3D
parallelepipedic, the data being represented in the tube inspected, amplitude possibly being an extra dimension.
FIG. 6 is a schematic longitudinal sectional view of a system formed a sensor, its water column and tube, and with different illustration trips ultrasound forming echoes. It helps to understand the complexity of these journeys, and the difficulty of the analysis.
Figure 6A is a schematic amplitude / time diagram of the signal ultrasound at level of a sensor working at oblique incidence. From the moment Texcit of excitation of the sensor, we find a water-tube interface echo at the moment Tinterf (that we can also note TphiExter0). Are then marked (dashed line) vertical) the instant TphiInter where the ultrasound beam reaches the inner skin of the tube, where he stands reflects and refracts, as well as the instant TphiExterl where the beam ultrasound reaches the outer skin of the tube. Due to oblique incidence, there is no echo reflexive significant that returns to the sensor in TphiInter in the absence of imperfection to this place.
This also applies in TphiExterl.
Figure 6B is a schematic amplitude / time diagram of the signal ultrasound at level of a sensor that works under normal incidence. Chronology general signals is the same as for Figure 6A (within a factor, related to incidence). By against, under normal incidence, there are significant echoes in TphiInter and in TphiExterl, even in the absence of imperfection at the relevant places of the tube.
Currently, non destructive testing systems used in production of tubes operate by making the ratio K between:
the amplitude As of a signal coming from the tube to be inspected, and the amplitude A0 of the signal originating from a reference standard defect, for the type of considered control. This standard reference defect is generally defined on a tube standard equipped with an artificial defect (for example a U-shaped or V-notch) chosen dimensional characteristics, for example in accordance with a standard of non-destructive testing, and / or the specifications of a customer.
The implicit assumption is that this signal amplitude is proportional to criticality imperfection, that is to say at its depth (DD). The graph of Figure 7 (good known to those skilled in the art, see Nondestructive Testing Handbook - chapter statistics of volume 7 published by ASNT - American Society for Nondestructive Testing) represents the actual distribution K = f (DD). It shows that in reality, correlation is very bad (of the order of 0.3 to 0.4 for ultrasonic testing).
More precisely, on the graph of FIG. 7, if the amplitude of reference AO
(K = 1) on the value XL (depth of the maximum acceptable imperfection) at center distribution (itself based on the oblique TDis), we see that we can again find imperfections at K = 0.5 DD depth greater than XL. he follows that, as a precaution, we have to fix AO for a value that is significantly lower than XL. By therefore, there is no production of tubes that would, in fact, be satisfactory. It is all the more harmful, economically, that the techniques of tubes are still quite heavy, both in terms of complexity and in terms of in energy.
The Claimant has therefore endeavored to improve the situation.
Figure 8 shows an improved device compared to that of Figure 4.
The output of the amplifier 73 is applied to a stage 761, which digitizes the amplitude of the signal from the amplifier 73, and works on this digitized signal. This treatment will be described below with reference to FIG.
floors 764 and 766 functionally similar to those in Figure 4. The raw signal of the sensor, such that visible on the oscilloscope 750, is called A-Scan by the men of the job. he includes echoes according to the scheme defined in Figure 6.
It is desirable to proceed to an imaging of imperfections of the tube, help from ultrasound signals. We will now describe obtaining an image.
In practice, an image is obtained by considering several explorations successive tube by a Px sensor, at successive angles that substantially cover a straight section of the tube. It is possible to do it with successive shots by one sensor, using the relative rotation tube / sensor.

Here, by way of non-limiting example, in the case of a installation of the kind said with a turning head.
In FIG. 8A, a Px sensor is considered, which can be of one of the types Pl, fold, P12, P21 and P22 above. In the example shown, this sensor Px comprises in does n elementary sensors Px-1, ..., Px-i, Px-n, which are aligned along the axis longitudinal of the tube, and which are fired at the same time. On the Figure 8A, this which is between the elementary sensors and the 3D graph of output 769 can to be considered as a converter.
The Ascan signal of the first elementary sensor Px-1 is applied to a amplifier 73-1, followed by two parallel tracks: that of the selector 763-1A and that of the selector 763-1B. Each selector 763-1A includes two outputs of amplitude respectively maximum and flight time. The maximum amplitude output is connected to a digitizer line 765-1Aa. The flight time output is connected to a digitizer of line 765-1At.
The maximum amplitude 765-1Aa line digitizer output is connected to a storage 768-Aa data buffer collecting data from the digitizers of line of maximum amplitude 765-iAa of index i ranging from 1 to n. The sottie of digitizer line 765-1At flight time is connected to a data buffer storage 768-Has collecting data from flight timeline scanners 765-IAT
of index i ranging from 1 to n. The line scanner output 765-1Ba amplitude maximum is connected to a 768-Ba data buffer storage data from line scanners of maximum amplitude 765-iBa of index i ranging from 1 to n. The 765-1Bt line-of-flight digitizer output is connected to a storage 768-Bt data buffer collecting data from the digitizers of line 765-iBt flying time of index i ranging from 1 to n.
On the basis of the information obtained during the passage of the standard tube, the operator can enter in the buffer storage 768-Aa and 768-At the information T_lA
corresponding to an indication of position and temporal width, which designates, in function of the known geometry of the tube, the moments when he will find an echo of skin internal, relative to the interior of the tube, for example the first echo Int1 of FIG.
The figure 6A shows more clearly the corresponding Int temporal window, around of TphiInter.
Similarly, on the basis of the information obtained during the passage of the standard tube, the operator can enter the 768-Ba and 768-Bt buffers T_1B information corresponding to an indication of position and temporal width, which means, in function of the known geometry of the tube, the moments when he will find a skin echo external, relative to the outside of the tube, for example the first echo Exil of Figure 6.
Figure 6A shows more clearly the time window Ext correspondingly, around TphiExter.
The diagram is repeated for the other Px-2 sensors, ... Px-i, Px-n.
Thus, each time selector 763 defines time windows kept from the moment of emission of the ultrasounds, and the intervals of time predeterminable where one can expect echoes about this selector. The illustration of figures 6 watch how can one define the interesting time intervals, taking into account from the angle ultrasound beam on the tube, as well as the diameter (internal or external) and the thickness of the tube. A given time interval corresponds to an echo given in a given point of the tube, for a given relative position between the tube and the sensor.
For simplicity, we admit here that the moments of shooting are synchronized on the rotation relative tube / sensors, so that an elementary sensor always works on the same longitudinal generator of the tube. The output of its selector therefore provides a after spaced out from analog signal samples, which each correspond to amplitude an expected echo on a wall of the tube. These Px-1 sensor samples (by example) are scanned in 765.
Synchronism with transmission can be provided by a link (no represented) with the transmitter 70, or with its trigger, the synchronization circuit 753, or its base of time 752 (FIG. 8). The display 750 can be maintained, if desired.
The system can operate on a tube rotating at a substantially constant speed. In this case, the angular velocity and tube advance can be measured using a encoder accurate angle, for example the model RS0550168 provided by the company Hengstler, and a laser velocimeter, for example the model LSV 065 provided by the company Polytec. The tube may also not turn, while it is the sensor system that turned. In in this case, the laser velocimeter is sufficient to measure the advance of the tube, while the speed of The rotation of the sensors can be known by means of an angular encoder.
For a given shot, the set of sensors Px-1 to Px-n provides a line of one picture, which corresponds to a cross section of the tube. In the other dimension of the image, a given elemental sensor provides a line that corresponds to a generator of the tube.
The scanners 765-1Aa, 765-2Aa, ..., 765-iAa, 765-nAa and 765-1At, 765-2At, ..., 765-iAt, 765-nAt allow to fill an internal image, relative to the skin internal of the tube. The scanners 765-1Ba, 765-2Ba, ..., 765-iBa, 765-nBa and 765-1Bt, 765-2Bt, ..., 765-iBt, 765-nBt allow to fill an external image, relative to the outer skin of the tube, with Tvol max the flight time of the echo amplitude Max.
The parallelepipedal 3D graph stored in 769 is for the sensor or group of Px sensors considered. Each point of this image corresponds, transposed into shading of gray, has an amplitude value of the echo due to the reflection of the signal ultrasound on a possible imperfection of the tube area considered. This value can also represent the ratio between the maximum amplitude of the ultrasound signal picked up on the tube during the test and the maximum amplitude of the ultrasonic signal obtained with a fault artificial reference standard, as defined above. The 3D graph cuboid is a representation of the preparatory Bscan 3D digitized in 769 - preparatory in this meaning that it serves as a basis for the generation of the Bscan 3D tube. The shape of the graph 3D is generally distinct from the form of the product under examination, in particular for tubes.

The information of the parallelepipedal 3D graph can comprise the set of the pairs (flight time, amplitude) of the AScan curve over a period of scan determined.
Parallelepiped 3D graphs scanned in 769 include 3D graphs parallelepipeds 891 constructed from data from a group of collectors and parallelepipedic graphs 892 constructed from data from a group of sensors P12 and P21 and P22 respectively as represent in Figure 11.
This image now corresponds to an area of the tube, obtained by the meeting of the substantially annular areas of the tube that correspond to each of the lines digitized. In fact, these are annular or helical zones if the ultrasound beam is applied substantially perpendicular to the axis of the tube. We know he is differently depending on the relative movement tube / sensor. The zones are then rather elliptical and, in fact, left or "twisted" in space. In this description, the expression "annular zones" covers these different possibilities.
It should be noted that to obtain a complete reconstruction of the 3D graph, information additional positioning of the sensor relative to the tube is necessary. She is available as a separate input 740. This information comes from an encoder or a together lasers for measuring spatial positioning. As the tube can to be assimilated to a cylinder without thickness, the position information can be reduced to two dimensions.
It is understood that the implementation of the invention on an existing bench of control by ultrasound ("UT bench") implies:
- accessibility to raw ultrasonic test data ("UT Raw Data "), which for example using an acquisition card, such as the NI 6024 model E series or National Instrument NI 6251 Series M, or by direct access to data digital bench control electronics, - the availability of online information on the speed of rotation (from tube or sensor) or the relative angular position of the tube relative to the sensor, and - the availability of online information on the advance speed of the tube or position linear relative of the projected sensor on the axis.
The diagram of Figure 8A can be applied:
in parallel with a P1 type sensor and a P12 type sensor, which observe the same area of the tube in two different directions. Each sensor is going to permit to obtain an internal image and an external image. Then one of the pictures maybe chosen according to a command noted Int / Ext.
in parallel with a P21 type sensor and a P22 type sensor, which there too, go each allow to obtain an internal image and an external image.
The diagram of FIG. 8A can also be applied to a sensor Pi, in which case three parallel channels are provided behind each amplifier (at least virtually).
One of the channels works on a repetitive time slot positioned as indicated in Volum. in Figure 6B. This way allows a control of imperfections in volume, that is to say in the thickness of the tube.
The other two channels can operate respectively on the slots time repetitive positioned as shown in WphiExter0 and WphiInterl on the Figure 6B. These two other ways allow a measurement of the thickness of the tube.
The distinction between the 3 channels is purely functional (virtual). In effect, said two other paths can be physically the same, in which one discriminates against moments or slots WphiExter0 and WphiInterl. We can also use only one physical way, in which one discriminates the moments or crenels WphiExter0, Volum. and WphiInterl.
It is representative to describe in more detail the case of a P11 type sensor with a P12 type sensor. That's what we will do now.

It is recalled that these two groups of P1 and P12 sensors are used for detection longitudinal imperfections on the tubes. Ultrasonic testing is realized with ultrasound shots (US) in two preferred directions ("dock wise" -"counter dock wise "):
- A sensor or group of Pli sensors provides an ultrasonic image of the tube in work direction ("dock wise").
- A second sensor or group of P12 sensors provides an ultrasound image of same tube in another direction of work ("counter dock wise").
Thus, the longitudinal imperfections are advantageously detected with 2 sensors or groups of sensors whose beam axes are symmetrically inclined by relative to a plane perpendicular to the axis of the tube. The inclination is by example about +/- 17. This provides an example of two-system application sensors, or two groups of sensors, as mentioned above.
In the embodiment of FIG. 8B, each scanning window 782 issue an amplifier 781 can be characterized by a start, a duration and a frequency which define a number n of points of the AScan signal taken in consideration. Each scanning window 782 then provides a number n of couples information (Amplitude, Flight Time), for each ultrasound firing. The Buffer / Multiplexer 788 delivers all the data thus collected in the graph 3D parallelepiped 769 taking into account the respective positions of sensors at moment the signal was received, this both thanks to the knowledge of the configuration geometry of the sensors relative to each other, and thanks to information from tube / sensor positioning at the time of the ultrasonic firing 740.
Reference is now made to Figure 9. For the first meaning of control (tab direction 1 selected), the images 903 and 904 are sectional views (respectively transverse and longitudinal) of the Bscan 3D Tube, 3D with the geometry of the tube, such as described later, from the P11 sensors. The positioning of these cuts is fixed thanks to the cross-sectional parameters at (mm) and longitudinal section at (degrees) . The images 905 (internal) and 906 (external) are CScans, such that defined previously, the image 905 (respectively 906) focusing on an area Ascan's temporal pattern, with internal skin imperfections (respectively external) are supposed to be detected. The information necessary for the reconstruction of 905 pictures and 906 are from the parallelepiped 3D BScan 891 of FIG.
The image 901 is a 3D representation in transparency of the Bscan 3D Tube of a portion of the product to be tested, a portion in which zones are identified potentially interesting, as described below. The same images 903 bis, 904 bis, 905bis, 906 bis and 902 are reconstituted for the second sense of control (tab meaning 2 activated), see Figure 9A.
We recall here that the foregoing description relates to the detection of defects to longitudinal orientation. The same approach applies to the search for defaults transversal (with sensor groups P21 and P22).
Reference is now made to Figure 11. Image blocks 901 and 902 are obtained from parallelepiped 3D graphs 891 and 892 using the block of transformation 930 as detailed in FIG. 11A. The 891 converter block of the figure 11 corresponds to the assembly of FIG. 8A, applied to the sensor P11. Similarly, block converter 892 also corresponds to the assembly of FIG. 8A, but applied to P12 sensor. The 891 and 892 converter blocks use the data from context tube / sensors in block 740. These data relate to the characteristics of the tube in exam courses and sensors in use.
The transformation block 930 is arranged downstream of the 3 D graphs cuboid 891 and 892 and may have the structure shown in Fig. 11A. The block of transformation 930 performs a time calculation of the propagation path waves in the tube taking into account the conversion of modes at the moment of impact a ultrasonic wave on a fault. At impact, a transversal wave can become longitudinal wave and vice versa. The transformation block 930 can estimate the propagation of the energy of the acoustic beam from calculations of coefficients of transmission and reflection. An analysis of the frequency spectrum of Ascan may be performed. The transformation block 930 may include a database real or simulated tests for comparison with 3D graphs received. The transformation block 930 fear restore the image Bscan 3D with the geometry of the tube.
As illustrated in FIG. 11A, the transformation block 930 comprises two blocks 931 and 932 Eliminate Areas of 3D Bscans Not Useful from a Graph 3D, the block 931 processing the data of 3D images 891 and block 932 processing the data from 3D images 892, two blocks 933 and 934 filtering by application of a window simulated time, respectively downstream of blocks 931 and 932, a block of simulation theoretical 935, a tolerance calculation block 937 feeding a block algorithm inverse 936, block 936 providing the images 901 and 902 defined previously.
The elimination by blocks 931 and 932 makes it possible to reduce the quantity processed information keeping the potentially interesting areas to represent so three-dimensional. Filtering can be done in length from a Cscan.
The length chosen may be greater than the length of a zone of amplitude greater than a threshold.
The 3D parallelepipedic Bscans can then be treated, including a zone potential imperfection.
The filtering by the blocks 933 and 934 can be carried out by confining the window time by the interface and background echoes. These filter blocks can also to limit the angular area of the potentially interesting tube and if necessary shift these areas in order to identify and completely reconstruct the potentially interesting. The images from blocks 933 and 934 are reduced Bscan 3D.
The theoretical simulation block 935 may include a database of simulations, for example of Ascans or Bscans 3D depending on the types and position faults. The database may include simulated results and / or of the test results on natural and / or artificial defects. The block reverse algorithm 936 can compare theoretical Ascans or Bscans 3D from the block of theoretical simulation 935 and 3D Ascans or Bscans obtained during the inspection so to determine the closest theoretical Ascan or Bscan 3D and, by therefore, the default (s) most likely. For example, the algorithm block inverse 936 compare a filtered experimental Ascan corresponding to a position in length and at a angular position with the theoretical Ascans on this same position in length and in developed. As another example, the inverse algorithm block 936 compares a Bscan 3D resulting from a reduced 3D Bscan corresponding to a position in length with the Bscans 3D theoretical on this same position in length. Both comparisons can to be carried out. The best set of theoretical representations of echoes is then the set with the smallest deviation from the data experimental.
After the transformation block 930, the filters 921 and 922 are illustrated, see Figure 11, which allow in particular to make extracts of the images, and their data from preparation, as input data gathered by the combiner block 960 for the neuronal or expert treatment 970.
In the embodiment described, the filter 921 has:
a signal output Zcur designating a working zone in the image.
This exit is used by an extraction function 951 which consequently performs a extracted from the image (Cscan) for Zcur zone, and access to the preparation 891 to obtain stored information (so-called Ascan) relative to the same area Zcur. All of this data is transmitted by the function extraction 951 to the combiner 960, as inputs to the neural processing or expert 970, an output providing information obtained by filtering, some at least Zcur zone, which it transmits as input of the neuronal treatment or expert, - optional (dashed line) complementary data outputs filtered to a memory 990.
It is the same for the filter 922, with the extraction function 952, for the same Zcur current zone.

The neural system 970 feeds a logic of decision and alarm 992, which pilot an automatic sorting and marking 994. It can be provided an interface interpretation 996 by an operator, who may present all or part of the data contained in the memory 990, in relation to the portion of tube under examination. The data contained in the memory 990 come from filters 921 and 922.
In addition to its prediction (origin, type and severity of indication) the system neuronal 970 provides an assessment of the confidence that can be placed in this prediction.
This information is accessible to operators who also have data more qualitative alternatives such as the order history in course or the problems that occurred during product development.
The operator, or a specialist can then intervene to weight the predictions in result.
Here, Figure 11 deals with information from at least two groups of sensors providing the same function or intended for the same type of control (the 2 groups Pll and P12 or both groups P21 and P22). The same scheme can be used to treat the information from a larger number of sensor groups intended to Has different type controls. The number of images processed at the same time is increases especially.
The primary function of filters 921 and 922 is to determine areas imperfections in the images 901 and 902. In general, the filtering is arranged to locate the areas to be analyzed, and to distinguish imperfections from other indications. The filtering works on two counterparts of two images. Both filters can work together.
By scanning the digital image, the locations of the image are first located or there are potential imperfections. For this purpose, it is possible to apply a threshold fixed established by calibration.

A threshold that adapts to the current noise level can be used the image. The method is based on the theory of detecting a signal in a noise white, which can be based on two assumptions:
HO hypothesis: measure = mean white noise m_b and standard deviation std_b Hypothesis H1: measurement = signal + white noise Statistical tests are done to determine if one is in the framework of the hypothesis HO, or hypothesis Hl. These statistical calculations are performed in real time on n slippery points of the image corresponding to shots consecutive, the number n can be determined by learning.
According to this method (case called "Gaussian additive"), it is possible for example to use the criterion of Neyman-Pearson to determine a detection threshold based on a probability of false alarm (pfa) given. This is expressed by the formula [21] attached. We use the Gaussian cumulative function, usually called Q (or the function Error erf), which must be reversed to obtain the threshold, according to the formula [22]
attached.
In practice, there is frequently the presence of background noise that can to have several origins (for example: presence of water inside the tube, snoring electrical, acoustic phenomena due to the structure of the product material control).
The use of a variable threshold avoids the false alarms that occur if we apply a fixed threshold.
Among the other false indications that may appear, parasites himself manifest by very brief peaks in the ultrasound signal. These parasites can be discarded by simple algorithms that we can call algorithms of count cumulative or integrators (example: "n blows before alarm" or "double threshold").
The plaintiff again considered the "turn", which is the path followed by the sensor the along the cylindrical surface to which the tube is assimilated. Filtering may be performed along each turn to further reduce the rate of false alarms. We use for this purpose for example a Butterworth filter, and / or a Fourier transform discrete, such as a fast Fourier transform. This technique is applied to each digital line.
The same kind of algorithm can be applied in the direction of the length of the tube.
Thus, potential imperfections are localized. When an imperfection is spotted, its position corresponds to the position analyzed in the images of the Figure 9 (by example), with a 3D image, a cross section and an axial section. The indications of radial position / thickness (or simply of situation internal, external or bulk imperfection) can be represented as attributes points of the image. We will thus have:
- two 2D images representing possible skin imperfections external tube, - two 2D images representing possible skin imperfections internal tube, and a 2D image representing the possible imperfections in the thickness of the tube.
Imperfections confirmed now, after elimination of parasites and false alarms, in particular.
For the rest, the Applicant has currently chosen to work on a image area fixed size. It is necessary to frame this zone on the existence data imperfection that we just got.
In other words, it is necessary to position the points identified as being higher than threshold to determine the complete area around an imperfection. It's a need by example, if one wishes to determine the obliquity of an imperfection.
The algorithm is structured around different stages:
- contour detection (Roberts gradient for example), - dilation (gathering of close contours), - erosion, then closing, which allows to determine a mask around imperfections - a last step of surrounding allows to locate completely imperfection.
For each imperfection, we thus obtain the coordinates of the image area corresponding, which will be useful for neural network analysis which intervenes then.
Figure 12 illustrates this treatment of the image areas in the form of a diagram of flux.
At the beginning of the images (801), there are zero to p image areas to be treated, such as with a confirmed imperfection. Operation 803 assumes that there is at least one first zone, which serves as the current zone to treat Zcur in 805. For this zone Zcur operation 807 selectively extracts the data from images 901 and 902 which correspond to this area (defined by its coordinates in the image).
operation 809 selectively extracts data which is intervened in the preparing the images 901 and 902, which correspond to the zone Zcur. of the examples of these data will be given below.
operation 811 performs the neuronal or expert treatment itself, on which we will come back to.
- The results obtained for Zcur zone are memorized selectively in 813, in correspondence of a Zcur zone designation.
- The 820 test checks whether there is another zone to be treated in the image, in which case it starts again in 805 with this other zone as indicated in 821; otherwise the processing of the current image (s) is complete (822).
In the case of the processing of the sensor Pi, there is only one image, which change the number of input parameters. Apart from that, the treatment can be usually the even.

After the determination of each zone of interest Zcur, the filtering can include other functions. For these other functions, Figure 13 illustrates schematic the interaction between the filtering and the sequence of operations illustrated on the figure 11.
Figure 13 is similar to Figure 11, but only for image 901.
She does to appear:
the tube-sensor context elements of block 740, the extractor 951 which finds the data for zone Zcur, in the image 901 and his preparation 891, - an internal / external block 7410, indicating whether the imperfection in the zone Zcur considered is located in inner skin or outer skin.
What the filtering adds to the basic data, is defined in more details to know, for each zone Zcur (block 805), as indicated by the contents of the box dashed:
- a search for the angle of obliquity in 941, an indication of imperfection length 942, It can be added to this, in particular:
an indication of alignment in C-Scan, in 945, and - in 946, an indication on the existence of other imperfections in the same section right of the tube.
In the embodiment described, data such as 945 and 946 are around the memory 990. Other data goes to neural networks or systems experts 970. These are here separated into two functions, as will be seen now.
An imperfection in a tube can be defined by its position, its type, and its gravity often equated with its depth. In the embodiment described, the type and the degree depth of a tube imperfection are determined separately using of two neural processes of the same general structure, which will be detailed now on a example.

The case of the type of imperfection is treated according to Figure 14, while the case of gravity is treated according to figure 15.
The types can be defined for example as illustrated in FIGS. 10A
at 10D.
These figures illustrate four types, constituting a simplifying choice by report to the list of imperfections provided by the API that can be produced by process tube development. The titles in French and in English are those used by skilled in the art to describe the nature of imperfections. We observe that imperfections of types 1 and 3 are straight, those of figures 2 and 4 arcuate (to "chord").
A correspondence between the real imperfections and the four types above can to be defined as follows:
Name in English Name in English Assignment Notch Notch TYPE 1 Crack Tap Type 1 Straw / fold perpendicular or Seam (perpendicular) TYPE 1 right (rolling) Straw / fold (rolling) Seam (arcuate), "overlap" TYPE 2 Gravel Sliver TYPE 3 Original Rolled-in-slug billet TYPE 4 Gouge Stripe TYPE 4 Inclusion Inclusion TYPE 4 Lack of material ("defourned") Bore-slug TYPE 4 Overlapping / overlapping / folding Lap TYPE 4 Here, FIGS. 14 and 15 both utilize three-dimensional neural circuits neurons intermediates (or "hidden neurons"), denoted NC121 to NC123 for FIG.
and NC141 to NC143 for Figure 15.
Figures 14 and 15 have a number of inputs in common. To try to facilitate understanding, entries are illustrated by types of different traits.

The double lines indicate that the entries are multiple, that is to say repeated for each point of Zcur zone.
First of all, in 7410 it comes, in accordance with the state of selectors 763 concerned, information as to whether to treat an imperfection located in inner or outer skin of the tube wall. This information can also be obtained on the BScan 3D.
The second category of common input quantities includes the magnitudes of context, which come from block 740 (Figure 13):
- in 7401, WT / OD, which is the ratio of the wall thickness to the diameter of the tube, - in 7402, Freq, which is the working frequency of ultrasonic probes, in 7403, ProbDiam, which is the useful diameter of ultrasound probes.
The third category of common quantities includes quantities from the which can be considered as common to both sensors 921 and (or more). For example, the average of the results on the two sensors is or we take the most representative result (maximum / minimum, as the case may be) These quantities are the magnitudes in 9201, the obliquity of the defect, and in 9202 its length. These two quantities are easily identifiable in the two images of FIG. 9, who have a mirror symmetry.
Reference is now made to Figure 14 only. The category next magnitudes includes separate measurement quantities for each of the two sensors (or group of sensors), and for each zone Zcur, what is reflected on the drawing by the use of a double line.
For a first sensor, we have:
- in 9511, Ki, which is the ratio between the maximum amplitude of the signal ultrasonic encountered in Zcur zone and on image 901, in relation to the amplitude maximum reference standard defect mentioned above. In fact, in the example, the amplitude in each pixel of the image 901 is defined by this ratio; K1 is then simply the maximum amplitude encountered in the Zcur zone of the image 901; we note Pmaxl the point of the Zcur area where this maximum is met.
in 9512, QBE1 which is a quantity of the C-Scan called QuantBumpsEchodyn, representing the number of local maxima encountered in the Zcur zone of image 901 near the maximum amplitude point Pmaxl. This QBE1 number is limited to local maxima encountered in the vicinity of Pmaxl, on both sides, but without than the amplitude of the signal has fallen back below a corresponding level noise background. QBE1 will usually take either the value 1 or the value 2.
These two quantities come from the image 901, via the extractor 951, which reflects the notation 951 (901) in the drawing. It adds to it:
in 9518, RT1 which is a quantity representing the rise time of the echo in the native ultrasound signal says A-Scan, (this is the gap between when the signal is maximum and the last previous moment when the signal is at the noise level of background, commonly expressed in microseconds). This size RT1 has been previously measured at the output of the amplifier 73 concerned (FIG. 8A); she was stored, by example in 891, in correspondence of the point of the tube which it concerns. It is as well as can be recovered selectively by the extractor 951. The size RT1 can now be directly measured by the operator on the 903 image of the Figure 9, or still on the parallelepiped 3D BScan.
For the second sensor, we have:
- in 9521, K2, which is defined as Ki, but for the image 902 instead of the image 901.
In the example, K2 is simply the maximum amplitude encountered in the zoned Zcur of image 902; we note Pmax2 the point of the Zcur zone where this maximum is meet.
- in 9522, QBE2 is defined as QBE1, but in image 902 instead of the image 901, and in the vicinity of Pmax2. Again, QBE2 will usually take either value 1, which is the value 2.
These two quantities come from the image 902, via the extractor 952.
adds:

in 9528, RT2 which is a quantity representing the rise time of the echo in the native signal called A-Scan. As before, this size RT2 has been earlier measured at the output of the amplifier 73 concerned (FIG. 8A); she was stored, by example in 892, in correspondence of the point of the tube which it concerns. It is as well as can be recovered selectively by extractor 952. The size RT2 can now be directly measured by the operator on the 903A image of the Figure 9, or still on the parallelepiped 3D BScan.
The last input 958 of the neural network is a constant value, noted Constanta, which represents a constant determined during the calibration of the model and resulting of learning.
The output 998 of FIG. 14 is a quantity indicative of the type of imperfection and its average inclination (defined according to the type).
The case of the degree of depth (or gravity) of the imperfection is treated according to the figure 15. The entries are the same as for Figure 14, except:
for the first sensor, the block 9512 is replaced by a block 9513, which treat a size EW_1, or EchodynWidth, which is the width at half height (50%) of the Echodynarnic waveform, for this first sensor. This magnitude EW_l is from the Cscan.
- similarly, for the second sensor, block 9522 is replaced by a block 9523, which treats the size EW_2, or EchodynWidth, which is the width at half height (50%) the echodynarnic waveform, for this second sensor.
- in 959, the constant, noted now ConstantB, is different.
- the output 999 is an indication of severity of imperfection, denoted DD.
It is noted that in both cases (FIGS. 14 and 15), a neural circuit 970 given deals an image extract 951 for one of the ultrasonic sensor groups, as well an extract 952 corresponding to the same area, but from another group of sensors.

The Applicant has observed that it is possible to obtain very good results.
satisfactory subject to a proper adjustment of the parameters of an expert system, by examples of neural circuits, and possibly their number, for optimize the prediction.
In addition, the plaintiff has found that by combining information collected by the different neural networks it was possible to further refine the prediction.
Overall, the input parameters of the neural network or the system expert are then features of the two 3D images (ratio of the max amplitude by report to amplitude of standard, echo width, representative echo orientation of the obliquity of the imperfection ...) and the control (sensor, dimensions of the tube...).
Output parameters are the characteristics of the imperfection (depth, tilt / type). The decision and / or alarm 992 can be taken automatically to using selected decision criteria, based on thresholds, with a margin of security as required. To define these thresholds, we can use the results of learning.
Reference is now made to Figure 16, which is a model of the circuit neuronal elementary of Figures 14 or 15, for two sensors.
This model includes an input level or layer IL ("Input Layer"), which regroup all input parameters (often called "input neurons"). Not to not overload the figure, are represented only three neurons El to E3, plus one constant, which can be considered as an EO neuron. This constant is most often called bias. In practice, input neurons are more numerous, according to Figure 14 or Figure 15, as appropriate.
Then at least one level or layer HL ("Hidden Layer" or "layer") is provided hidden "), which includes k neurons (only 2 of which are represented to not overload the drawing).

Finally comes the output neuron Si, which provides the decision, in the form a value of representative of the importance of an imperfection of the tube, for example a longitudinal imperfection. This output corresponds to block 998 in the figure 14 and 999 in Figure 15.
Note that the "neuron" - constant E0 intervenes to weight no only the or the HL hidden layers, but also the output neuron (OL layer or "Output Layer ").
The general behavior of a neural circuit as used here is given by the formula [11] of Annex 1, where wii is the weight assigned to the signal Xi present at the entrance of the neuron J.
In the circuit provided here, an elementary neuron behaves according to the formula [12]
as shown schematically in FIG. 17.
The output Si of FIG. 16 provides an estimated value which corresponds to the formula [13]
of Annex 1.
By training, the Applicant has adjusted the hidden neurons and their weights so that the function f is a nonlinear, continuous, differentiable function bounded.
The currently preferred example is the arc-tangent function.
We know that a neural network determines its coefficients wii commonly called synapses by learning. This learning must involve typically 3 to 10 times more examples than there is weight to calculate, while covering correctly the beach desired working conditions.
Starting from examples Ep (p = 1 to M), we determine for each example the difference Dp enter here value Sp given by the neural circuit and the actual value Rp measured or defined experimentally. This is recalled by the formula [14].

The performance of the neural circuit is defined by a magnitude overall difference Cg, called "cost". It can express itself for example according to the formula [15], as an overall magnitude of weighted quadratic difference.
Learning poses different problems in a case like control of the imperfections in the tubes, in particular because they are heavy, as already indicated.
The plaintiff first conducted a first apprenticeship on simulation. We can use for this purpose the CIVA software developed and marketed by the Office of Atomic Energy, France. This first learning allowed of identify the influential parameters, and build a first version of the network of neurons based on virtual imperfections. The cost function has been optimized.
The applicant then conducted a second apprenticeship combining results obtained on simulation and artificial imperfections, that is to say created intentionally on real tubes. This second learning allows to build a second version of the neural network, whose cost function has also been optimized.
The Applicant then combined the results obtained with imperfections artificial, and on a set of imperfections present on tubes real, these imperfections being accurately known by posterior measurements out of production line. This third phase validated the last version of neural network. This version proved to be operational for the monitoring manufacturing. However, when it is installed on a new installation or modified, it is now appropriate to subject him to a "calibration", using a dozen artificial samples covering the entire range of imperfections treat. he naturally follows an optimization.
Figures 11, 12, 14 and 15 have been described in the context of the sensors Pli and P12.

The same principle can be applied to the sensor group Pl. In this case, it will not be there no picture 2, and the built network has fewer input parameters, like already indicated. The circuits described for two sensors can be used for a alone but without input parameters for the part Image 2.
The same principle can also be applied to both sensor groups P21 and responsible for detecting transverse imperfections, taking into account the fact that sensors are for this detection inclined (for example +/- 17) in a passing plan by the axis of the tube.
It will be understood that, in each case, there is a digital processing of the type defined by Fig. 11, elements 992 to 996 excepted. This treatment is overall designated by 763, according to Figure 8, where it is followed by blocks 764 and 766.
An assembly as shown in FIG. 18 is thus obtained, with:
for the sensor P1, a treatment 763-1, followed by a decision phase and alarm 764-1;
- for the P1 and P12 sensors, a 763-10 treatment, followed by a decision and alarm 764-10;
- for the P21 and P22 sensors, a 763-20 treatment, followed by a decision and alarm 764-20;
- the three phases 764-1, 764-10 and 764-20 being interpreted together by the sorting and alarm automaton 767.
A variant of FIG. 18, not shown, consists in providing only one single phase Decision & alarm, using directly the outputs of the three treatments 763-1, 763-and 763-20.
Nondestructive testing proper is done "on the fly", ie as and as the tube passes through the control system. The decision treatment the information described above can be taken either as and when as the ube scrolls in the control installation (with decision-alarm and marking "to the "variant" is to make that decision after the whole length of tube was inspected, or even later (after checking the whole lot of tubes for example), each tube being identified / identified (N of order by example).
In this case, it is necessary that the information obtained be recorded (Stored). Records can be analyzed posterior by a operator empowered to make a decision after analyzing the results recorded and treated by the network (s) of neurons.
Of course, given the properties of neural circuits, it is possible to at least partially group together all the neuronal circuits (contents in the treatments 763-1, 763-10 and 763-20) in a single neural circuit, having all the entrees required.
The described embodiment directly uses neural networks to title example of expert systems. The invention is not limited to this kind of production.
Here, the term "neural circuit arrangement" may cover other techniques nonlinear data analysis, with or without neural circuits.
In general, the converter may comprise an amplitude input maximum in a selector and a corresponding flight time entry. said entries can provide sufficient data for the compliance or non-compliance decision conformity of a product.
The transformation block may include a data elimination element unnecessary, a filtered area filter element, a simulator and a element interpretation. Reducing the amount of information allows a speed of higher treatment.
The simulator may include a theoretical simulation element, a calculator tolerance and an inverse algorithm.
The output stage can include:

a combiner arranged to prepare digital circuit inputs neuronal, to from an extract of the images corresponding to a zone of imperfection presumed properties of the presumed imperfection in the same zone, resulting from the filter, and of data context, at least one neural circuit, which receives inputs from the combiner, - a digital decision and alarm stage, operating on the basis of the output of the circuit neuronal, and an automatic sorting and marking machine, arranged to separate and mark tubes decided not compliant by the digital decision and alarm stage.
The system proposed here has been described in the case of non-destructive testing when manufacture of seamless tubes, in which case the invention applies particularly good. The same techniques can be applied in particular to products long ironworks not necessarily tubular.
In the case of welded tubes or other welded products (such as sheets or plates), the system proves to be able to determine in addition the limits of the cord of welding, and therefore locate any imperfections in the cord of welding, which may be to watch. For their part, imperfections located outside weld bead boundaries, which may be inclusions already present in the basic strip (or product) are to be considered differently.

Annex.
Section 1 (11) = F (EE w + Ivo) (12) S = + (13) Dp = Sp ¨ (14) EP = m D2 Cg = p = 1 p (15) Section 2 1 _______________________________ (threshold ¨ 7711, pfa = e 'b dx = Q ____________________ (21) Him / -ViTr stdb stdb threshold = stdb Q-1 (P fa) + Mb (22)

Claims (16)

42
1. Operational assistance tool device, for non-operational control destructive, in progress or at the end of production, steel products including at least one tube or other long products, this tool being intended to extract information on potential imperfections of the product, from return signals that capture (73), consecutive to the selective excitation (70) of ultrasonic sensors transmitters according to a chosen law of time, ultrasonic receiver sensors forming a selected geometry arrangement, mounted in ultrasonic coupling with the product by through a liquid medium, with relative movement of rotation / translation between the tube and the arrangement of transducers, said operating tool comprising:
a converter (891; 892) capable of selectively isolating a representation number of possible echoes in designated time windows, function relative rotation / translation movement, said representation comprising the amplitude and the flight time of at least one echo, and to generate a 3D graph cuboid a transformation block (930) capable of generating a 3D image (901; 902) possible imperfections in the tube from the 3D graph and a base of data, a filter (921; 922) capable of determining, in the images (901; 902), areas of presumed imperfection (Zcur), as well as properties of each alleged imperfection, and an output stage configured to generate a compliance signal or not conformity of a product, wherein the transformation block comprises an elimination element of useless data, a filtered area filter element, a simulator and a element of interpretation.
2. Device according to claim 1, wherein the converter (891;
892) includes a maximum amplitude input in a selector and an input of time corresponding flight.
3. Device according to claim 1, wherein the simulator comprises a element theoretical simulation, a tolerance calculator and an inverse algorithm.
4. Device according to any one of claims 1 to 3, wherein the floor of output includes:
a combiner (960), arranged to prepare digital inputs of from an extract (951; 952) of images corresponding to a zone of presumed imperfection (Zcur), and properties of the alleged imperfection in the same area, from the filter (921; 922), at least one neural circuit arrangement (970), which receives inputs from work from the combiner (960), a digital decision and alarm stage (992) operating on the basis of the exit the arrangement of the neural circuit (970), and a sorting and marking automaton (994) arranged to separate and mark decided non-compliant products by the digital decision and alarm stage (992).
5. Device according to claim 4, wherein said operating tool comprises two converters (891,892) devoted respectively to two arrangements of ultrasonic transducers (P11, P12, P21, P22) of selected geometry (P11, P12, P21, P22), mounted in ultrasonic coupling substantially in accordance with symmetry mirror the direction of their respective ultrasonic beams, and the combiner (960) is arranged to operate selectively on the internal skin echoes or on the echoes outer skin or echoes occurring in the mass of the tube, but even time on the data relating to one and the other of the two arrangements transducers.
6. Device according to claim 4, wherein the converter (891; 892) is arranged to selectively isolate a numerical representation of possible maximums echoes in designated time windows corresponding to echoes of skin internally, with echoes of external skin, as well as echoes from the mass of tube, respectively, and the combiner (960) is arranged to operate selectively on the echoes of internal skin or on the echoes of external skin or on the echoes intervened in the mass.
7. Device according to claim 4, wherein the combiner (960) receives at least an input (9511; 9521) relating to an extremum of amplitude of the image in the zoned presumed imperfection.
The device of claim 4, wherein the filter (921,922) is arranged for produce, as properties of each alleged imperfection, its obliquity and its length, while the combiner (960) receives corresponding entries imperfection obliquity (931) and imperfection length (932).
9. Device according to claim 4, wherein the filter (921; 922), the combiner (960), the neural circuit arrangement (970) and the digital decision stage and alarm (992) are arranged to operate iteratively on a series of zones presumed imperfection (Zcur) determined by said filter (921; 922).
The device of claim 9, wherein the filter (921; 922), the combiner (960), the neural circuit arrangement (970) and the digital decision stage and alarm (992) are arranged to operate alternately on the inner skin and the skin external tube.
Apparatus according to claim 4, wherein said arrangement of neural circuit includes:
a first neuronal circuit (NC121-NC123) able to evaluate the nature of a imperfection among a plurality of predefined classes, and a second neural circuit (NC141-NC143) capable of evaluating the severity of a imperfection.
12. Device according to claim 11, wherein the two circuits neurons have entries that differ by:
an entry (9512; 9522) of a number of neighboring maxima for the first circuit neuronal, and an echo width input (9513; 9523) for the second neural circuit.
13. Device according to any one of claims 11 and 12, wherein the Outputs from both neural circuits are combined to refine the prediction.
14. Device according to any one of claims 1 to 13, wherein the issue and the reception of the ultrasonic signals are carried out each time by a even transducer, for at least part of the sensor arrangement.
15. Non-destructive testing device for tubes in or out of manufacturing, comprising:
an arrangement of ultrasonic transducers of selected geometry mounted in ultrasonic coupling with the tube via a coupling medium, with relative movement of rotation / translation between the tube and the arrangement of transducers circuits for selectively energizing (70) these transducer elements according to a law of chosen time, and to collect (73) the return signals they capture, and an operating aid tool according to any one of claims 1 to 14.
16. Non-destructive testing of iron and steel products including minus one tube or other long products, in process or in production, including following steps :
at. provide an arrangement of ultrasonic transducers of selected geometry, mounted in ultrasonic coupling with the tube via a medium coupling, with relative movement of rotation / translation between the tube and the arrangement of transducers b. selectively exciting (70) these transducer elements according to a law of time selected, vs. collect (73) the return signals they capture, in order to analyze selectively these return signals (760-766), to derive information sure possible imperfections of the tube, said information comprising the amplitude and the flight time of at least one echo, and generate a parallelepiped 3D graph, d. selectively isolate a numerical representation of possible echoes in designated time windows, depending on the relative movement of rotation / translation (891; 892), and draw a 3D image (901; 902) imperfections possible in the tube from the 3D graph and a database, and e. generate a signal of conformity or nonconformity of a product wherein step e., comprises:
e1. filtering (921; 922) images (901; 902) according to filtering criteria choose, to determine areas of presumed imperfection (Zcur), as well as properties of each alleged imperfection, e2. form (960) numerical work inputs, from an extract (951 ;
952) images corresponding to an area of presumed imperfection (Zcur), properties of the presumed imperfection in the same zone, resulting from the filter (921 ; 922), and context data (740), e3. apply the inputs thus formed (960) to at least one arrangement of neural circuit (970), e4. numerically processing the output of the neural circuit arrangement (970) according to selected decision criteria, to draw a decision and / or alarm (992), and e5. remove and mark (994) the tubes decided non-compliant by step e4.
CA2709611A 2007-12-21 2008-12-16 Non-destructive testing, in particular for tubes during manufacture or in the finished state Active CA2709611C (en)

Priority Applications (3)

Application Number Priority Date Filing Date Title
FR07/09045 2007-12-21
FR0709045A FR2925690B1 (en) 2007-12-21 2007-12-21 Non-destructive control, especially for tubes during manufacturing or in the final state.
PCT/FR2008/001751 WO2009106711A2 (en) 2007-12-21 2008-12-16 Non-destructive testing, in particular for tubes during manufacture or in the finished state

Publications (2)

Publication Number Publication Date
CA2709611A1 CA2709611A1 (en) 2009-09-03
CA2709611C true CA2709611C (en) 2017-06-06

Family

ID=39636907

Family Applications (1)

Application Number Title Priority Date Filing Date
CA2709611A Active CA2709611C (en) 2007-12-21 2008-12-16 Non-destructive testing, in particular for tubes during manufacture or in the finished state

Country Status (17)

Country Link
US (1) US8365603B2 (en)
EP (1) EP2223098B1 (en)
JP (1) JP5595281B2 (en)
KR (1) KR101476749B1 (en)
CN (1) CN101903771B (en)
AR (1) AR069323A1 (en)
AU (1) AU2008351946B2 (en)
BR (1) BRPI0821312B1 (en)
CA (1) CA2709611C (en)
EA (1) EA021646B1 (en)
ES (1) ES2616552T3 (en)
FR (1) FR2925690B1 (en)
MY (1) MY152235A (en)
SA (1) SA3046B1 (en)
UA (1) UA99844C2 (en)
WO (1) WO2009106711A2 (en)
ZA (1) ZA201004112B (en)

Families Citing this family (50)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
FR2903187B1 (en) 2006-06-30 2008-09-26 Setval Sarl Non-destructive control, especially for tubes during manufacturing or in the final state
US9104195B2 (en) 2006-12-20 2015-08-11 Lincoln Global, Inc. Welding job sequencer
US9280913B2 (en) 2009-07-10 2016-03-08 Lincoln Global, Inc. Systems and methods providing enhanced education and training in a virtual reality environment
US9318026B2 (en) 2008-08-21 2016-04-19 Lincoln Global, Inc. Systems and methods providing an enhanced user experience in a real-time simulated virtual reality welding environment
US9196169B2 (en) 2008-08-21 2015-11-24 Lincoln Global, Inc. Importing and analyzing external data using a virtual reality welding system
US8911237B2 (en) 2008-08-21 2014-12-16 Lincoln Global, Inc. Virtual reality pipe welding simulator and setup
US9483959B2 (en) 2008-08-21 2016-11-01 Lincoln Global, Inc. Welding simulator
US8834168B2 (en) 2008-08-21 2014-09-16 Lincoln Global, Inc. System and method providing combined virtual reality arc welding and three-dimensional (3D) viewing
US9330575B2 (en) 2008-08-21 2016-05-03 Lincoln Global, Inc. Tablet-based welding simulator
US8657605B2 (en) * 2009-07-10 2014-02-25 Lincoln Global, Inc. Virtual testing and inspection of a virtual weldment
US9011154B2 (en) 2009-07-10 2015-04-21 Lincoln Global, Inc. Virtual welding system
US8747116B2 (en) 2008-08-21 2014-06-10 Lincoln Global, Inc. System and method providing arc welding training in a real-time simulated virtual reality environment using real-time weld puddle feedback
US8851896B2 (en) 2008-08-21 2014-10-07 Lincoln Global, Inc. Virtual reality GTAW and pipe welding simulator and setup
JP5694193B2 (en) * 2009-01-19 2015-04-01 ゼテック インコーポレイテッドZetec,Inc. Automatic nondestructive test analysis method of eddy current
US8274013B2 (en) 2009-03-09 2012-09-25 Lincoln Global, Inc. System for tracking and analyzing welding activity
US9773429B2 (en) 2009-07-08 2017-09-26 Lincoln Global, Inc. System and method for manual welder training
US9221117B2 (en) 2009-07-08 2015-12-29 Lincoln Global, Inc. System for characterizing manual welding operations
US20160093233A1 (en) 2012-07-06 2016-03-31 Lincoln Global, Inc. System for characterizing manual welding operations on pipe and other curved structures
US9230449B2 (en) 2009-07-08 2016-01-05 Lincoln Global, Inc. Welding training system
CN106233358A (en) 2014-06-02 2016-12-14 林肯环球股份有限公司 System and method for artificial welders training
US8569655B2 (en) 2009-10-13 2013-10-29 Lincoln Global, Inc. Welding helmet with integral user interface
DE202009014770U1 (en) * 2009-11-02 2011-07-04 Seuthe, Ulrich, 58300 Device for checking a component for damage
US8569646B2 (en) 2009-11-13 2013-10-29 Lincoln Global, Inc. Systems, methods, and apparatuses for monitoring weld quality
US9468988B2 (en) 2009-11-13 2016-10-18 Lincoln Global, Inc. Systems, methods, and apparatuses for monitoring weld quality
US8884177B2 (en) 2009-11-13 2014-11-11 Lincoln Global, Inc. Systems, methods, and apparatuses for monitoring weld quality
CN101788534A (en) * 2010-03-24 2010-07-28 中国石油集团渤海石油装备制造有限公司 Method for detecting transverse defect of submerged-arc welding seam
US8798940B2 (en) * 2010-04-16 2014-08-05 Olympus Ndt Inc. Rotating array probe system for non-destructive testing
WO2011139142A1 (en) * 2010-05-03 2011-11-10 Röntgen Technische Dienst B.V. A method for inspecting an object by means of ultrasound
FR2960960B1 (en) * 2010-06-03 2012-07-20 V & M France Method and device for assistance in controlling the production of tube traverability
FR2974437B1 (en) * 2011-04-21 2013-10-25 Eads Europ Aeronautic Defence Method for simulation of non-destructive control operations in real conditions using synthetic signals
US9020766B2 (en) * 2011-09-23 2015-04-28 Mastinc. Multi-modal fluid condition sensor platform and system therefor
US9389215B2 (en) * 2011-09-23 2016-07-12 Mastinc Multi-modal fluid condition sensor platform and system thereof
US8991258B2 (en) * 2012-05-10 2015-03-31 General Electric Company Linear scanner with rotating coupling fluid
US9767712B2 (en) 2012-07-10 2017-09-19 Lincoln Global, Inc. Virtual reality pipe welding simulator and setup
US9568461B2 (en) 2012-12-31 2017-02-14 Mastinc Multi-modal fluid condition sensor platform and system therefor
US9207639B2 (en) * 2013-01-24 2015-12-08 General Electric Company Transforming A-scan data samples into a three-dimensional space for facilitating visualization of flaws
GB2512835A (en) 2013-04-08 2014-10-15 Permasense Ltd Ultrasonic detection of a change in a surface of a wall
DE102013106901A1 (en) * 2013-07-01 2015-01-08 Bundesrepublik Deutschland, vertreten durch das Bundesministerium für Wirtschaft und Technologie, dieses vertreten durch den Präsidenten der BAM, Bundesanstalt für Materialforschung und -prüfung Apparatus and method for detecting material defects in rotationally symmetrical test specimens by means of ultrasound
JP6230841B2 (en) * 2013-07-26 2017-11-15 旭化成エンジニアリング株式会社 Pipe thinning evaluation method using wall thickness measuring device
US20150072323A1 (en) 2013-09-11 2015-03-12 Lincoln Global, Inc. Learning management system for a real-time simulated virtual reality welding training environment
US10083627B2 (en) 2013-11-05 2018-09-25 Lincoln Global, Inc. Virtual reality and real welding training system and method
US9836987B2 (en) 2014-02-14 2017-12-05 Lincoln Global, Inc. Virtual reality pipe welding simulator and setup
CN103926317A (en) * 2014-04-25 2014-07-16 上海势华电子科技有限公司 Method for screening workpieces through ultrasonic wave
KR101722063B1 (en) 2015-06-24 2017-04-21 경상대학교산학협력단 Detection system for extremely small defect of ultrasonic lock in infrared thermography
KR101736641B1 (en) * 2015-12-24 2017-05-17 주식회사 포스코 An apparatus and a method for detecting a crack
US10684261B2 (en) 2016-04-01 2020-06-16 General Electric Company Ultrasonic bar and tube end testing with linear axis robot
EP3319066A1 (en) 2016-11-04 2018-05-09 Lincoln Global, Inc. Magnetic frequency selection for electromagnetic position tracking
RU2655048C1 (en) * 2017-06-21 2018-05-23 Общество с ограниченной ответственностью "Нординкрафт Санкт-Петербург" Device for ultrasonic examination of round stock and pipes
RU2685744C1 (en) * 2018-04-23 2019-04-23 Родион Николаевич Юрьев Method of decoding defectograms and digitized signals of investigation of solid bodies
BR102018015331A2 (en) * 2018-07-26 2020-02-04 Vallourec Solucoes Tubulares Do Brasil S A method for assessing the inclusional level in steel tubes using high frequency transducer for automatic ultrasonic inspection

Family Cites Families (21)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4625557B2 (en) * 1985-02-20 1998-05-26 Rutherford Lora E Acoustical imaging systems
JPS61286747A (en) * 1985-06-14 1986-12-17 Hitachi Ltd Three-dimensional probe and three-dimensional imaging apparatus
US4803638A (en) * 1986-06-26 1989-02-07 Westinghouse Electric Corp. Ultrasonic signal processing system including a flaw gate
KR930009636B1 (en) * 1990-08-24 1993-10-08 이재범 Filtering method on the transform domain having the papeline structure
JP3007474B2 (en) * 1991-04-19 2000-02-07 中部電力株式会社 Ultrasonic inspection method and apparatus
CN1109594A (en) * 1994-04-22 1995-10-04 清华大学 Acoustic coupling method and apparatus for sound emission detection
JPH09171006A (en) * 1995-12-20 1997-06-30 Ishikawajima Harima Heavy Ind Co Ltd Device or evaluating ultrasonic flow detection data
JPH10115604A (en) * 1996-10-14 1998-05-06 Ishikawajima Harima Heavy Ind Co Ltd Ultrasonic flaw detection evaluating device
JPH10311138A (en) * 1997-05-14 1998-11-24 Ohbayashi Corp Construction for identifying single pipe
JP3730429B2 (en) * 1999-01-29 2006-01-05 株式会社日立製作所 Ultrasonic flaw detection result display method and ultrasonic flaw detection apparatus
FR2796153B1 (en) * 1999-07-09 2001-11-30 Setval Non-destructive control with distributed ultrasonic sensors
EP1305594B1 (en) * 2000-05-30 2010-01-06 Oyo Corporation, U.S.A. Apparatus and method for detecting defects or damage inside a sewer pipeline
JP2002257802A (en) * 2001-02-27 2002-09-11 Mitsubishi Heavy Ind Ltd Apparatus for visualizing ultrasonic signal
FR2833706B1 (en) * 2001-12-13 2004-07-23 Setval Non-destructive control with ultrasonic sensors of metallurgy products
JP4167841B2 (en) * 2002-03-22 2008-10-22 クレノートン株式会社 Intelligent ultrasonic flaw detection system using neural network
JP4115954B2 (en) * 2004-02-27 2008-07-09 株式会社東芝 Ultrasonic inspection equipment
JP4542813B2 (en) * 2004-04-26 2010-09-15 株式会社東芝 3D ultrasonic inspection equipment
JP2006255083A (en) * 2005-03-16 2006-09-28 Ge Medical Systems Global Technology Co Llc Ultrasonic image formation method and ultrasonic diagnostic equipment
JP4728762B2 (en) * 2005-10-03 2011-07-20 東芝プラントシステム株式会社 Ultrasonic flaw detection image processing device
JP4859521B2 (en) * 2006-05-02 2012-01-25 三菱重工業株式会社 Program, processing apparatus and processing method for processing ultrasonic flaw detection data
FR2903187B1 (en) 2006-06-30 2008-09-26 Setval Sarl Non-destructive control, especially for tubes during manufacturing or in the final state

Also Published As

Publication number Publication date
KR101476749B1 (en) 2014-12-26
SA3046B1 (en) 2013-02-02
ZA201004112B (en) 2011-02-23
CA2709611A1 (en) 2009-09-03
WO2009106711A3 (en) 2009-10-29
ES2616552T3 (en) 2017-06-13
EA021646B1 (en) 2015-08-31
AU2008351946A1 (en) 2009-09-03
CN101903771A (en) 2010-12-01
FR2925690B1 (en) 2010-01-01
WO2009106711A2 (en) 2009-09-03
JP2011506992A (en) 2011-03-03
AR069323A1 (en) 2010-01-13
BRPI0821312A2 (en) 2015-06-16
BRPI0821312B1 (en) 2019-04-30
FR2925690A1 (en) 2009-06-26
MY152235A (en) 2014-09-15
EA201070787A1 (en) 2011-04-29
US8365603B2 (en) 2013-02-05
AU2008351946B2 (en) 2014-01-09
EP2223098B1 (en) 2016-11-23
UA99844C2 (en) 2012-10-10
EP2223098A2 (en) 2010-09-01
JP5595281B2 (en) 2014-09-24
KR20100110340A (en) 2010-10-12
US20100307249A1 (en) 2010-12-09
CN101903771B (en) 2013-08-14

Similar Documents

Publication Publication Date Title
CN101477085B (en) Three-dimensional ultrasonic imaging device
NL1025267C2 (en) Method and device for examining the internal material of the object from a surface of an object such as a pipeline or a human body with the aid of ultrasound.
EP1701157B1 (en) Eddy current inspection method and system using multifrequency excitation and multifrequency phase analysis
US7997138B2 (en) Method for inspection of metal tubular goods
US9177371B2 (en) Non-destructive examination data visualization and analysis
US7255007B2 (en) Configurations and methods for ultrasound time of flight diffraction analysis
RU2453836C2 (en) Method and device for automatic ndt of tubular wheel axles with variable inner and outer radii
JP5113340B2 (en) Method and system for inspecting an object using ultrasonic scanning data
US7480574B2 (en) Method to characterize material using mathematical propagation models and ultrasonic signal
Carvalho et al. Reliability of non-destructive test techniques in the inspection of pipelines used in the oil industry
EP0106580B1 (en) Acoustic detection of defects in structures
Komura et al. Crack detection and sizing technique by ultrasonic and electromagnetic methods
US7995829B2 (en) Method and apparatus for inspecting components
JP4096014B2 (en) Ultrasonic inspection method and apparatus for reactor pressure vessel
JP4854288B2 (en) Method and system for ultrasonic inspection
EP2703806B1 (en) Non-destructive evaluation methods for aerospace components
RU2498292C1 (en) Method and apparatus for ultrasonic flaw detection
US6591679B2 (en) Method for sizing surface breaking discontinuities with ultrasonic imaging
Ostachowicz et al. 50th anniversary article: comparison studies of full wavefield signal processing for crack detection
EP2508879B1 (en) 3d ultrasonographic device
CN106796204A (en) System for checking track with phased-array ultrasonic
US10324066B1 (en) System and method for the improved analysis of ultrasonic weld data
Iyer et al. Evaluation of ultrasonic inspection and imaging systems for concrete pipes
EP2906906A1 (en) A method of locating and sizing fatigue cracks
US7606445B2 (en) Methods and systems for ultrasound inspection

Legal Events

Date Code Title Description
EEER Examination request

Effective date: 20130320